Hormone therapy and breast cancer: conflicting evidence
Cindy FarquharCochrane Menstrual Disorders
and Subfertility Group
The world of hormone therapy in the 1990’s
Throughout the 1970s, 1980s and 1990s long term use of HRT was widely recommended for women after the menopause
Suggested Benefits: prevention of hot flushes, osteoporosis, heart disease, ageing, improved cognitionPossible harms: breast cancer, venous thromboembolism
Balancing the benefits and harms
-12
-10
-8
-6
-4
-2
0
2
HeartDisease
HipFracture
Est+Prog
Dea
ths
indu
ced-
prev
ente
d / 1
000
wom
en tr
eate
d w
ith H
RT
Stroke
Other Breast Cancer
Daly et al 1996
HRT and Breast Cancer is not a new story……
Berquivist (1992) RR of breast cancer in HRT users 1.6Nurses Health study (1995) RR 1.45 after >5 years use of HRTLancet Collaborative meta analysis (1997) RR 1.35 after >5 years of HRT
Lancet Collaborative GroupMeta analysis of 51 observational studies on breast cancer risk and HRT use52,000 women with breast cancerMajority on ERTAdjusted for age of menopauseMain findings:
RR of breast cancer diagnosis 1.35 after 5 + years of HRTNo increased risk in past users (>5 years)Risk greater in slim women than overweight (BMI >25) womenFamily history of breast cancer did not increase riskNo increased risk of mortalityNo difference between ERT and HRT
Design of WHI studyPOPULATION:
16,608 women aged 50 to 79 yearsPopulation based sample (recruited from mailing and media awareness)Heterogeneous group – minimal exclusionsWashout period before trialTwo study groups: HRT and ERT
What sort of women were in the trial?Age: 50-79 years with mean of 63 years70% overweight, 45% BMI ≥ 30Ethnically diverse20% prior HRT use, 6% current users
1667-61610Colorectal cancer
2000-51510Fracture
550+82129DVT
1250+71326Stroke
1430+73037Heart disease
1250+83038Breast cancer
NNT for 1 additional event
Additional events/1,000
placeboHRT
WHI results for HRT: July 2002Cases per 10,000 women per year
No overall increase in death rates at 5.2 years of follow up
Prior use of HRT increased risk cf with no prior use of HRT
HR 2.13 (1.15-3.94) for prior useHR 1.06 (0.81-1.38) for no prior use
Adherent to therapy increased risk
HR 1.49
ERT only study: 2004
No effect on BC diagnoses reported after 6.8 years follow up
HR 0.77 (nominal 95%CI 0.59-1.01 and adjusted 95% CI 0.57-1.06)
Mortality from BC: no difference but no HR provided
Million Women Study – Aug 031996-2001: National Health Service Breast Screening Programme invited women to take part prior to entry1084110 women, 50-64 years50% of women had used HRT18 % had BMI ≥ 30
Million Women StudyRelative Risk
E only 1.3 (1.21-1.40) E+P 2.0 (1.88-2.12)
Current users of HRT cf never usersDiagnosis of BC: adj RR 1.66 (1.58-1.75) Mortality from BC: 1.22 (1.00-1.48)
Past users of HRT: no increased riskNo differences: formulations and delivery systems
Mortality from BC in MWS
Authors conclusions
10 years of HRT is estimated to result in5 additional cancers per 1000 users of E only
19 additional cancers per 1000 users of E+PUse of HRT by women 50-64 yrs in UK in past decade has resulted in an estimated 20,000 extra breast cancers (15,000 from E + P)
Summary of WHI and MWS: RR
4.1 years4.1 years6.8 years5.6 yearsTime frame
Not reported
1.22 (1.05-1.41)
Not reported
0.95 (0.24-38.1)
BC mortality
1.3 (1.21-1.40)
2.0 (1.88-2.12)
0.79 (0.61-1.02)
1.26 (1.02-1.56)
Br Ca ∆
MWS – E only
MWS –E+P
WHI – E only
WHI –E+P
Cochrane Review
WHI and MWS: consistency?
Not reportedNot reportedBC mortality E only
IncreasedNo differenceBC ∆ E only
Increased No differenceBC mortalityE+P
IncreasedIncreasedBC ∆ E+P
MWSWHI
Possible explanation for the differences
Study designRCT versus observational
Power16,000 women versus 1,000,000
US vs UK popDifferences in screening etc
Prior use of hormones72% E+P (WHI), 52% E (WHI) versus 50% in MWS
Younger age in MWSMean 63yrs E+P (WHI), 63 E (WHI) versus 56 years (MWS)
BMIWomen in WHI study heavier than women in MWS
Explanation for the differences: BMI amongst Estrogen only in MWS
45%18%
37%
45%
MWS
34%
21%
WHI 2004
0.99 (0.73-1.34)
≥ 30
1.14 (0.94-1.40)
25-29
1.36 (1.14-1.63)
< 25
RR of BC ∆MWS
BMIKg/m2
Weight and breast cancer
Overweight women have increased risk of breastIn MWS trial women with BMI < 25 had increased risk
ConclusionsAuthors of trial concluded that:
HRT should not be used for long-term disease prevention because the benefits were not sufficient to justify the risks. On balance the harm of HRT was greater than the benefit (global index)The trial was not designed to assess the effects of HRT for short term use to control menopausal symptoms
Intervention/comparisonCombined HRT study
Conjugated equine oestrogens 0.625mg/day + medroxyprogesterone acetate 2.5mg/day in 1 tablet Placebo tablet, 1 tabletParticipants and study staff blinded but unblindingoccurred because of need to treat bleeding
Estrogen only studyWomen who had undergone hysterectomyConjugated equine oestrogens 0.625mg/day Placebo tablet
Time period of trialRecruitment from 1993 – 1998Average follow up 5.2 yearsPlanned duration 8.5 years (until 2005)Trial stopped early because:
Test statistic for invasive breast cancer exceeded the stopping boundaryGlobal index statistic supported risks exceeding benefits
OutcomesPrimary
CHD rates – HRT expected to be a benefitInvasive breast cancer rates – HRT expected to be a harm
Other outcomesHip fracture and other fracture ratesStroke ratesVTE ratesEndometrial cancer ratesColorectal cancer ratesTotal death ratesGlobal index
Lancet Editorial Dec 2004“But the HRT story is an all too familiar one in modern
medicine. A new drug is found to be potentially useful in a large proportion of the population. Hypotheses for extended use, in the case of HRT to prevent cardiovascular disease and bone fractures, are generated from observational studies. Its use is then heavily promoted beyond the initial indication. Rigorously conducted randomised studies with long enough follow-up are scarce or lacking. Harm and risk are uncovered many years after widespread use. “